Phenomenology of B -> pi pi, pi K Decays at O(alpha^2 beta_0) in QCD Factorization

Phenomenology of B — > tttt, ttK Decays at O{a 2 s (5o) in QCD

Factorization

Craig N. Burrell
Department of Physics, University of Toronto
60 St. George Street, Toronto, Ontario, Canada M5S 1A%

Alexander R. Williamson

Department of Physics, Carnegie Mellon University,
Pittsburgh, PA 1521%

We study 0(a 2 Po) perturbative corrections to matrix elements entering two-body
exclusive decays of the form B — ► tttt, ttK in the QCD factorization formalism, includ-
ing chirally enhanced power corrections, and discuss the effect of these corrections
on direct CP asymmetries, which receive their first contribution at 0(a s ). We find
that the O(a 2 f3o) corrections are often as large as the 0(a s ) corrections. We find
large uncertainties due to renormalization scale dependence as well as poor knowl-
edge of the non-perturbative parameters. We assess the effect of the perturbative
corrections on the direct CP violation parameters of B° — ► 7r + 7r~.

2

I. INTRODUCTION

In recent years a wealth of new data on two-body nonleptonic B decays to light pseu-
doscalar final states has beenproduced by the CLEO 0,33, BABAR 0, 0,0,011110
Q El and BELLE Il4 , Il5l llfiL Il7{ | experiments. This experimental program provides

a rich context for the precision study of the weak sector of the standard model. However,
in non-leptonic decays all of the final state particles are QCD bound states which interact
strongly with one another. There is therefore nonperturbative physics in the low energy
matrix elements which is an obstacle to precision calculations.

These low energy matrix elements can be evaluated if it is assumed that they factorize
into simpler matrix elements For example,

{K + K~\(ub)v-A(su) V - A \B) -> (K-\(su) V - A \U)(-K + \(ub) V - A \B).

(1)

The matrix elements on the right-hand side can be parametrized in terms of form factors
and decay constants. This 'naive factorization' prescription has in some cases proven to
be a remarkably successful approximation 0, 21], EjJ- As it stands, however, there is no
way to improve the calculation by making systematic corrections in a controlled expansion.
Moreover, the missing 'non-factorizable' physics is responsible for final-state rescattering
and strong interaction phase shifts, and is therefore of considerable interest.

Beneke, Buchalla, Neubert, and Saehrajda (BENS) 000 have argued that for
certain classes of two-body nonleptonic B decays the str ong interactions which break fac-

torization are perturbative in the heavy quark limit [22J, |23(. The physical picture behind
this claim is 'color transparency': gluons must be energetic to resolve the small color dipole
structure of the energetic light meson in the final state. The BBNS proposal, called QCD fac-
torization, was accompanied by a demonstration that, for heavy-light final states, it holds up
to two-loop order j^l in the heavy quar k limit. This conclusion was subsequently extended
to all orders in perturbation theory [2I El •

The BBNS proposal reproduces naive factorization as the leading term in an expansion in
a s and Aqcd/^6; thereby placing naive factorization on a more secure theoretical foundation.
Although there has been some recent progress

2J

El El BBSS;

at present there

exists no systematic way to address the Aqcd /vrib corrections. The perturbative corrections,

on the other hand, can be calcu
decay modes

22, 123

2J,|3J,|35,

ated, and the 0(a s ) corrections are known for a variety of

In this paper we study 0(ol 2 s (3q) perturbative corrections to B decays of the form B — >
7T7T, ttK. Though this is only a subset of the full 0(a 2 ) correction, the method is motivated
by the empirical observation that the O{a 2 (3o) contribution often dominates the full result.
For example, this is true for R(e + e~ — > hadrons) T(r — > v T + hadrons) j^l, and

r(6 — > X u ev e

The dominance of the O{a 2 s (3o) contribution becomes rigorous in the

very formal 'large-/?o limit' of QCD, where the number of colors is fixed and the number of
flavours rij — > —00, resulting in /3 = 11 — 2/3rif — > 00.

3

The perturbative corrections we consider arise from three sources: 'non-f actor izable' ver-
tex corrections, QCD penguin diagrams, and spectator quark interactions. In the BBNS
framework the computation of these amplitudes requires the introduction of a number of
nonperturbative parameters. We study the numerical significance of the uncertainties due
to these parameters. We neglect power corrections of the form 0(AQCD/m,b) n , with the
exception of a class of 'chirally enhanced' corrections that can be numerically significant.
Renormalon studies of these decays indicate that the leading power corrections from soft
gluons are at (9(AQco/' m fe) j^- This is in contrast to B — > Dir decays where a similar
analysis points to leading power corrections at 0(AQ CG /m b ) 2 |41, 42 1.

The calculations presented in this paper are similar to calculations performed earlier by
Neubert and Pecjak in j^J, although they differ in several important ways. The goal of
the previous paper was to study power corrections to B — > LL decays in a manner similar

to what had been done for B — > D^L decays in 4l|, |42|. The authors calculated the
amplitudes to O(l/flo), which is subleading in the large-/?o limit. With these expressions
they derived predictions for the CP asymmetries of several decay modes. The calculation
involved summing a class of graphs to all orders in perturbation theory, and extracting from
their large order perturbative behaviour information about power corrections. Estimates of
the O(a"/3o _1 ) corrections were also made in the large-/?o limit.

Our focus is the convergence behaviour of perturbation theory in the BBNS framework.
Instead of calculating to subleading order in the large-/?o limit, we restrict ourselves to
0(ol 2 s (3q) corrections. To avoid the need for Wilson coefficients evaluated at NNLO, we
concentrate on observables which vanish at leading order in perturbation theory. In partic-
ular, we study the direct CP asymmetries for six pseudo-scalar final states Aqp(tiii, ttK).
Though several of these modes also exhibit indirect CP violation, we restrict our discussion
to direct CP violation only. We find that the O(a 2 (3o) corrections are similar in size to the
0(a s ) values. We also find that the greatest uncertainty in the CP asymmetries is due to
the renormalization scale dependence, which is enhanced by the O(a 2 s f3o) corrections. In
contrast, the uncertainties induced in the asymmetries by nonperturbative parameters are
relatively small. Of particular interest is the mode Aqp(tt~ K°) which, to the order we work,
is independent of most of the nonperturbative parameters in the analysis.

he direct asymmetry
El- We find that

At the end of the paper we present a more detailed analysis of
parameter A nn , which has attracted considerable interest recently
this parameter receives a substantial correction at O(a 2 j3o), the size of which we give as a
function of the unitarity angle 7. We examine the relationship between this quantity and the
current experimental values, and find them to be in agreement within the large theoretical
and experimental uncertainties.

The structure of this paper is as follows: in the first Section we briefly review the theoret-
ical context for our calculation; in the subsequent Section we give an outline of our method
and collect most of our analytical results. This is followed by a brief phenomenological
study of direct CP asymmetries for a variety of different decays to ixix, nK final states, pay-

i

ing special attention to the O{a 2 s (3o) corrections. In the concluding Section we summarize
our results.

II. THEORETICAL BACKGROUND

We work in the weak effective theory where the weak bosons and top quark have been
integrated out. The effective Hamiltonian, valid below My/, is

n oS = % E >® [ C i°i + + E + CsgOsg ] f h.c.

V £ p=u,c \ j=3 . 6

where the CKM matrix elements are X p = V* s V p b for the AS = 1 Hamiltonian and
X' p = V* d V p b for decays to non-strange final states. The effective operators mediating the de-
cays are divided into left-handed current- current operators (Oi 2 ), QCD penguin operators
(C3,...,6), and a chromomagnetic dipole operator (Os g ). Explicitly, the operator basis for the
AS = 1 Hamiltonian is |43[

0\ = (pb) V - A (sp)v-A, P 2 = (pib^V-A^jP^V-A,

03 = (sb) V _ A Eg (qq)v-A , 4 = (Sibj) V _ A Eg (WiW-A ,

5 = {sb)v- A E<? (QQ)v+A , O e = (Sib^v-A E? (WiW+A ,

8g = I^ m6 1^(1 + 75)^6. (3)

The AS = operator set is as above with the s fields replaced by d fields. In these
expressions we use the shorthand (qq')v±A = 97 M (1 ±75)9' f° r the Dirac structures. Roman
indices on quark fields denote SU(3) color structure, and the summations over q in the
penguin operators 3 _ 6 run over all five active quark flavours q G {d, u, s, c, 6}. Some
authors include in Q a set of electroweak penguin operators which, however, produce only
small effects and are neglected in our analysis. The values of the Wilson coefficients Cj(/x)
are obtained by matching the effective theory onto the full theory at fi = mw and running
down to /i ~ rrif,. This procedure has been carried out to NLO in QCD, the results of which
can be found in j^.

The low energy dynamics are contained in the matrix elements of the four-quark operators
Oi. In the BBNS framework these matrix elements are given by

(M l M 2 \0\B) = F B ^ Ah (m 2 M2 )f M2 jf* rfxT 7 (x)$ M2 (x) + (M 1 ~ M 2 )

+ [\xdydCT n (x,y,^MM^M 2 (x)^B(0 + o(^-) . (4)
Jo V m b J

where Mi is the meson which receives the spectator quark of the B meson and M 2 is called
the 'emission meson'. The nonperturbative elements in this expression are the B decay form
factors F B ^ M , the final state meson decay constants Jm, and the light-cone momentum

5

distribution amplitudes §Mi which give the probability for a valence quark to carry a par-
ticular fraction of the meson's light-cone momentum. The quark and antiquark composing
the emission meson M2 are assigned momentum fractions x and x, respectively. Likewise
the quark and antiquark in Mi are assigned momentum fractions y and y, respectively. The
light antiquark in the B meson is assigned momentum fraction £ of the B meson momentum.
In the heavy quark limit, we can neglect components of momentum transverse to the light
cone, and consider only the Fock state containing the valence quark and antiquark. Thus
we have x = 1 — x, and likewise for y and £.

The factorization-breaking corrections are contained in the hard-scattering kernels T J (x)
and T n (x,y,^), each of which has a perturbative expansion. At leading order, the hard-
scattering kernels take the values

22

23. 44]

x)

0{a s

T n (x,y,0 = + O(a s

and, given that the light-cone wavefunctions <J>Mi are normalized to unity,
naive factorization.

The nonperturbative light-cone distribution amplitudes (LCDAs) $m in
by 0,0

(5)

reduces to
are defined

(P(p)Mz2)q a (zi)\0)

(6)

.h

( J iXe i(.xp-z 2 +xp-z 1 )

fry 5 $(x) - /ip7 5 %{x)

' 6

a/3

In this equation is the meson twist-2 LCDA, and $ PjCr (a;) are twist-3 LCDAs which will
contribute to the 'chirally enhanced' power corrections below. The quantity fip appearing
in (jUJ) is a 'chiral enhancement' factor

= : (7)

m q + nig

where q and q are the quarks which comprise the valence state of the pseudoscalar meson
P. In practice one introduces these LCDAs into Feynman amplitudes by replacing quark
bilinears with a projection matrix M |24 |

up >a (xp)T )3 a ,abV a ,b{xp)

ifr

where

AN C Jo

dxM^pTfaaa

(8)

■%(x).

(9)

Pi -P2

In these expressions Greek indices denote Dirac structure, Roman indices denote color struc-
ture, and T is an arbitrary combination of Dirac and color matrices. The momenta pi^ are
the momenta of the meson quark and antiquark, respectively. In the collinear limit pi = xp
and P2 = xp. In order to arrive at from the definition (JBJ) the equations of motion for
the twist-3 LCDAs and an integration by parts have been used to eliminate $ CT 00.

6

Throughout this paper we write the twist-2 LCDAs as a decomposition over Gegenbauer

polynomials as

$m(x) = 6x(l - x)

5>f(//)Cf*(2x-l)

n=l

where the Gegenbauer polynomials CfJ 2 {y) are defined by the generating function

1 d n

c 3 J 2 (y)

n\ dh n

[1 - 2hy + hf

-3/2

(10)

(11)

h=0

The Gegenbauer moments af 4 have been studied using nonperturbative methods in QCD

49

3-

In the

and, for many light mesons, estimates exist for the leading moments af 2 (45 ,
far ultraviolet fi — > oo we have af 1 — > 0, so at the scale fi ~ mj, which is still large compared
to the nonperturbative scale of QCD, the Gegenbauer moments af 1 are expected to be
small. This statement will be made more quantitative in Section ITVl In the approximation
of including only 'chirally enhanced' twist-3 contributions, the twist-3 LCDA equations of
motion constrain $ p (x) to take its asymptotic form & p (x) = 1 j^.

Following the authors of j^l we use the factorization formula (J3J) to rewrite matrix
elements of (El) in the convenient form

(7iK\n eS \B)

^ £ \ p (kK\T p \B)

(12)

p=u,c

where

= ax 5 pu (ub) V - A ® (su)v-a
+ a 2 S pu {sb) V - A ® (uu) V -A
+ «3 Y^g (* b )v-A ® (qq)v-A
+ alJ2q(qb)v-A ® (sq)v-A
+ a5^ q (sb) v _ A (g) (qq) v +A

(13)

where {qq')s±p — ?(1 ±75)<7' ', and a summation over q G {u, d] is implied. There is a similar
expression for it it final states, obtained by replacing the s quark by a d quark. Matrix
elements of operators containing the <g> product are to be evaluated as one would in naive
factorization

(MxM 2 \jx ® j 2 \B) = (Mx\jx\B) (M 2 \j 2 \0) or (M 2 \j 1 \B) (Mx\j 2 \0)

(14)

where the choice depends on the specific quark content of the mesons in the process under
consideration. The nonfactorizable corrections are contained in the coefficients dj. Though
we have not explicitly indicated it in ()13j) . in general these coefficients are mode specific,
dependent on the shapes of the LCDAs of the final state particles. We present the explicit
forms for a, in section ILTT1

7

III. PERTURB ATIVE CORRECTIONS

We consider three classes of diagrams: factorization-breaking vertex diagrams, strong
interactions with the initial state spectator quark, and QCD penguin diagrams. They are
shown in Figures |2]-|U The first and third of these classes contribute to the hard scattering
kernel T 1 in (J3J); the second contributes to T 11 . In this paper we do not include t he p ower
suppressed weak annihilation diagrams which have been studied by other authors [241 ] .

The diagrams at (D(a 2 Po) are obtained by replacing the gluon in the 0(a s ) diagram by
a gluon with a fermion-loop self-energy correction, as shown in Figure ^ followed by the
replacement rij — > — 3/3 /2.

a\ (3 :

FIG. 1: The form of the dashed gluon line in Figures [2] - H] at each order in perturbation theory.
Where an undressed gluon line provides the 0(a s ) contribution, the fermion-loop self-energy
correction produces the 0(a 2 Po) contributions.

All of the perturbative corrections are contained in the coefficients <2j defined in (|13|).
These coefficients may be written as

ai = C x

a 4

9i

An ^ \ An

AC F n 2 J a s (n h )

N c

An

H

a 2 = C 2

Ci

M 2 M 1
2

An

47T V An

AC F n 2 I a s (/i h

as = C 3 +

9±

N c

l + C F {^V M2

An

Hm 2 m 1

An '"^ ' \ An
AC F n 2 \a 8 {nh)

N c

An

H

C 4 +

9i

M 2 M 1
2

\ An

PoVm 2

( Ots{Hh)

An

i + Cf {^v M2+ ^) p v M2

An 1X14 V An
AC F n 2 J a s (/i h )

An

H

M 2 Mi

An

M 2 M X

M 2 M X

M 2 Mx

PoH,

M 2 Mi

8

«5

N r .

C 6 +

9i

+

1-Cf

l + C F

M 2 ,2 h

AC F ir 2 \ a s (n h

4ir

47T

E 'm 2 m x

( _ 6) + m\ AM)

M 2 Mi

+ n: \-^r Pm °> 3+ l^rJ APj

A/2,3 f >

(15)

where Cj = Ci(fi). The nonperturbative physics is contained in the functions labelled V, H,
and P, according to the type of diagram from which they arise. In particular, the non-
factorizable vertex corrections, treated in section IIII Al below, produce the functions vj$
and Vm ■ The scattering of hard gluons off the spectator quark, treated in section UlI Bl
gives rise to the functions M and H^J 2Mi . Graphs with penguin topologies, discussed in
section IIII C[ produce the P%f n and P\ l2 n functions, where n refers to the twist of the LCDA
that enters the evaluation of the function. All of these functions consist of convolutions of
hard-scattering kernels with meson light-cone distribution amplitudes, as will be shown
below.

The scale at which renormalization scale dependent quantities are to be evaluated differs
depending on the source of the contribution. In particular, while the vertex and penguin di-
agrams are evaluated at a scale fi ~ m&, the spectator scattering contributions are evaluated
at a lower scale \Xh

v '771&Aqcd- This applies to all scale dependent quantities multiplying
the spectator scattering functions H^'\ including the Wilson coefficients

A. Vertex Diagrams

The first class of diagrams we consider are those shown in Figure 121 in which a hard
gluon is exchanged between the emission meson M2 and the quarks involved in the B — ► M\
transition. These amplitudes are proportional to both Jm 2 an d F^ Ml (0) and contribute to
the kernel T 1 in (J3J). In terms of the coefficients a; in (|15|) they produce vj^ 2 at 0(a s ) and
Vg at O(a 2 s f3 )-

Although when evaluated in D = 4 — 2 e dimensions each of these diagrams contains
a 1/e 2 pole from infrared and collinear divergences, these cancel in the sum of the four
diagrams, leaving a residual UV 1/e pole to be renormalized. We renormalize in the MS
scheme, treating 75 in the naive dimensional regularization (NDR) prescription j^lj. For

M,

M,

]

M,

FIG. 2: The factorization-breaking vertex diagrams,
completeness we restate the result of Ref. [3] for the 0(a s ) contributions

V

Al

6

5"

-1 +

/.

+ 3.

and

M

6

_ ln (ml)

5"

+ 114

l

+ 3.

dx <?(x)$m(x)

dx g(l - x)$ A/ (x)

where the integration kernel is

"3(1 -2x)

2(1 - x \ ~ 2 ^ + 3i?r ' ) + ( x ^ x )

+

ln(x)

2(1 -x]

For the next order result we find

2i7rln(x) — ln 2 (x) — 2Li2 (1 — x) — (x «-> x)

1 /

in 4 +-

n 2

+
65

In iL

(?(x)$m(x)

/ dx h(x)$ M (x) ~ *
/o 12

and

VL

M

111

dx g(l - x)$ M (x)

f 1 5
+ / dx h(l - x)$m(x) ~ —■
Jo 12

The function h(x) appearing in this expression is given by

h(x)

3(1 -3x)
' 4(1 -x)

'7(1 -2x) 3

tf( x ) + [ ' ^ + ^ ] ln(

4(1 -x) 2 y

x +

3xLi 2 (l — x)
2(1 -x)

10

+

4

5 — 3x

15 + 7in) + (x x)

21n(l -x) -in) ln 2 (x) +

4(1 -x)

4-3x
~ 2fl - a;'

4(1^) + 2 ^ - \") '" W

■Li 2 (l - x) - 2Li 3 (l - x) - 4Li 3 (x) - (x x)

(21)

This function has previously been derived in |42j; we confirm that result.

One may notice from a comparison of the factorized operator fT3*|) with the pattern of
vertex graph contributions to the coefficients in f!15|) that the unprimed functions V, V
are associated with (V — A) ® (V — A) operator structures, while V, V' are associated
with (V — A) <S> (V + A) structures. The remaining operator structure present in (fT3*|) is
(S — P) <E> (S + P), and this receives a nonzero vertex contribution only when the twist-3
LCDAs $ p are included. We have used the fact that, in the approximation of including only

-6' and

the 'chirally enhanced' terms at twist-3, Q p has its asymptotic form $ ?
the momentum fraction integrals, resulting in the constants
of fl"5j) at 0(a s ) and 0(a^Po), respectively.

Using the Gegenbauer expansion for the LCDAs $m
momentum fraction x to obtain

15

[x) = 1 to carry out
-4' appearing in qq

we carry out the integration over

dx g(x)&M(x)

3Z7T

11
~2~

3in a

M

21
20

a 2

+

79 2

171

36 3

and

(ix }i(x)&m(x) = n 2

33 K •
OZ7T

2

, 2 65 11.

37T H 171

6 2

a

a

M

(22)

+ I f

7359 9

400 10

27T « 9 + 7T

10481 37

720

- —in ) af +
15

(23)

Thus the vertex diagrams introduce complex phases into the amplitude, and the magni-
tude of the phase depends on the shape of the LCDAs parametrized by af 1 .

B. Spectator Scattering Diagrams

The two diagrams involving hard scattering with the spectator quark are shown in Figure

|3] The 0(a s ) corrections are expressed in Ref.
in (JT5J). We choose to write these functions as

24 in terms of two functions H

(')

M2M1

as shown

n M 2 M 1

^^)ll dx ll dy ll di

2fi

Mi

3H

(')

m 2 Mi

{x,y,{)® M2 (x)^(y)<S> B (t)

(24)

11

FIG. 3: The factorization-breaking spectator scattering diagrams.

where we have divided the integration kernel into twist-2 {^H^) and twist-3 ( 3 H^) com-
ponents. In the approximation £ <C x, y, which one expects to be valid through most of
phase space, the integration kernels are

1

M2M1

2H' M2Ml (x,y,£)

3H' M2Ml (x, y,
3 H M2Ml (x, y, f )

x y 6'
1

xyi'

(25)

If one replaces the light-cone distribution functions $m with their expansions in terms of
Gegenbauer polynomials (|1U|) and carries out the integrations in (124)) . one finds

H {,)

M

+

m B X B F
■X hh

B->M

6/ijv/i

-11

l (0)
1 =F «f 2

9(1

a

Mi

Mi

•)(l±af 2 + a 2 A/2 ±-

+ «f 2 T

•)

where the ellipses denote higher order Gegenbauer moments, and in ' db/
applies to and the bottom symbol to H'. Following Refs. 0,
introduced two parameters \ B and X^j 1 defined by

(26)

;he top symbol

241 ] we have also

1 , Mt)

m B

A

B

^ g (y)
y

yMi

(27)

Because the light quark in the B meson carries a small momentum fraction, the wavefunction
$b(£) has support only for < £ < Aq CD /m fe . The definition (|2*7j) then implies A# ~ Aqcd-
It is necessary to introduce the parameter X^ 1 because, with the asymptotic form for
&p !l = 1, the integration contains a logarithmic divergence when the B meson spectator
quark enters M\ as a soft quark y ~ 0. This divergence is a consequence of our having
neglected the small transverse components of momentum and quark off-shellness 24 1 . X^ 1
is therefore a new complex nonperturbative parameter, and by power counting it is of size

X^ 1 ~ ln(m b /A QC D)-

At next perturbative order one finds two new functions HjJ M defined by

(')

H

(')

M 2 Mi

In

m

B,

5
3

H

(')

M 2 A'h

12

A/i

m|F ^(0)^o

da; / dy d£

fV(')

{x,y,^M 2 (x)^ Ml (y)^B(0

+

2/i

Mi

Mi-

(28)

where the new hard scattering kernels are

2 H M2Ml (x,y,£) = 3 H' M2Mi (x,y^)
2H' M2Ml (x,y,£) = 3 H M2Ml (x,y,£)

Carrying out the integrations explicitly one arrives at

,2

Me y)

x y £

HZ V)

(29)

rL M 2 M 1

In

/r

m

B,

5
+ 3

H

(')

M 2 Mi

+

/b/mi

27

17

. J_(l +a M 1+a A/ 1 + .

43

-ft

Mi

9 1 ' 1J

Ma , _ Ma

1 ± ftf 2 + < 2 ±

M 2

X

Mi
ii

A,

X

Mi

A,

1 ± af 2 + ft£ J2 ±
1 =F «f 2

«f 2 T

(30)

The large coefficients in the second line of (j3*Uj) result from the integral Jq lay <&(y)/y dy.
In the Gegenbauer expansion of the LCDA $ all of the Gegenbauer moments on enter with
large coefficients, so only if the moments themselves decrease quickly will this integral be
well represented by the terms we retain. In addition we have been forced to introduce two
additional parameters similar to those in (|27|):

1 lnf $ fl (0 _ m B

[ d£
Jo

A,

C dy
Jo

ln£$f (y)

X

y

Mi
H ■

(31)

By power counting these parameters are of order X B ~ Aq CD / ln(Aq CD /m b ) and X^
Li 2 (-m 6 /A QCD ).

Mi

C. QCD Penguin Diagrams

An important source of strong phases in the decay amplitudes are the QCD penguin
diagrams, shown in Figure |U These diagrams give rise to the functions P M i and P M i
appearing in (|15j). The four quark operators in the Hamiltonian (J3J) contribute to the left-
hand diagram, while the chromomagnetic dipole operator Og g contributes in the right-hand
diagram.

13

We begin by stating the results at 0(a s ). At twist-2 one finds 24

pp

r M,2

jf (x) = a

dxP 2 p (x)$ A /(x)

4 m 6 2

-In — + - - G s p ,x

3 n 3

C 3

!l n !^ + l_ G( 0, x )_G(l,x)
3 /i 3

+ (c 4 + c (i

^± ln !^ _ ( n/ _ 2 )G(0, x) - G(s c , x) - G(l, x)

-2G

1 — x

(32)

where rif = 5 is the number of active quark flavours, s q = {m q /m b ) 2 , and Gg^ = C 8g + G 5 .

M 2 M 2

M, B

> i it i, ^ it it it

M,

FIG. 4: The penguin and magnetic dipole diagrams.

The function G(s, x) in (j32j) is given by the integral

G(s,x) — —4/ cfat w(l — u) ln[s — u(l — u)(l — x) — iel. (33)
Jo

The integral Jq 1 dx G(s p , x)$(x) is complex and contributes to the strong phase of the am-
plitude for s < 1/4; that is, for all quark flavours except the b quark.
The order 0{o 2 s (5q) results at twist-2 we find to be

- r M,2

mi

pp

M,2

dx ln(x

ie)PZ(x)$ M (x).

(34)

A similar situation exists when one turns to the twist-3 terms. One finds the leading
corrections involve

' l dxP$(x)<&™(x) (35)

pp

Gi

4 m& 2

-In — + - - G(s p , x)

6 fl 6

G s

^ In — + ^ - G(0, x) - G(l, x)
3 ji 3

(G 4 + G 6

i^L i n 1^ _ ( n 2 )G(0, x) - G( Sc , x) - G(l, x)
3 /i

-2G

where the twist-3 distribution function $^(x) has replaced the twist-2 distribution in (}32|) .
The O(a^P ) function is, in a manner closely analogous to (J3~4*j) . given by

pp

In

E

ml

+

pp

<ix ln(x

ie)P£(x)$f(x).

(36)

The expressions which result from carrying out the integrations over momentum fractions
are quite complicated for the penguin diagrams, and we refrain from presenting them here.

11

IV. PHENOMENOLOGICAL ANALYSIS

In this section we take the analytic results of section ITTT1 and study direct CP asymmetries
for various tttt and nK final states. We begin by giving the expressions for the decay
amplitudes and the definitions for the CP asymmetries. In the next subsection we collect
and discuss the input parameters we use to obtain numerical results. This is followed by a
presentation and discussion of the results.

A. Definitions of Branching Ratios and CP Asymmetries

the B

In terms of the coefficients a, and the factorized matrix elements defined by
ttK decay amplitudes are

.Gf , 2

z 7i ~ m

{ m M 2 ) JM 2 ,

(37)

A(B~ -

-V2A(B- -

-A(B° -
V2A(B°

7T~K

A,

p

2[i K
1

m b

-4,

K

a u Q>\ + Ap + Aj

TT+R-)

* 7r°K°)

m b
2fi K

a u a>\ ~r a„ CZ4 -\- a p a 6

V2A(B

AjrK + [A u CI2] Artt
A n K ,

--K )

7T

] K~)

IT

(38)

A(B- ->ti
— A(B°

In these expressions a, = a^K), \ p = V*V p b, and a summation over p e {u, c} is implicit
in expressions like X p af . The last relation is a consequence of isospin symmetry.
The B — > 7r7r decay amplitudes are given by

- .A(B°

7T 7T

A Tl

-V2A(B~ -> tt-tt ) = [A'Jai + aa)]^,

.4(£° -> TrV) = V^B" -> 7T-7r°) - ^(fi° -> TT+TT") , (39)

where now = ai(7T7r) and A' = VpbVL* d . The CP conjugate decay amplitudes are obtained
from the above by replacing X p ^ — > (aW) . Note that none of these decay modes are
dependent on 03 or 05. These factors, therefore, play no further role in our discussion.

CP violation can occur either directly via a difference between CP conjugate decay rates
(T(B — > f) ^ T(B — > f) or, for neutral B mesons, indirectly via B° — B° mixing. Accord-
ingly, we treat the two cases separately.

For the decays B~ — > tt~K°, B~ — > tt°K~ , and B° — > ir + K~ , the CP asymmetry is time
independent and is defined as

15

where our sign convention is set by defining B = B°, B~ as an initial state containing a b
quark, and B = B°,B + as containing an initial b antiquark. This CP asymmetry vanishes
in the limit of naive factorization, and first occurs at order a s in the BBNS formalism. As
such, it can be calculated to order a 2 /?o with knowledge of only the next-to-leading order
Wilson coefficients.

For the neutral B meson decays to final states / for which there are interference effects
between B° — > / and B° —>■ B° — > /, the resulting CP asymmetry is time dependent

A ( . f) _ \A(B°(t) ^ f)\* - \A(B°(t) ^ f)\*
cHlJ) \A(B°(t)^f)\* + \A(B°(t)^f)\2- 1 )

In this paper the modes which fall into this class are B° — > ir°Ks, B° — > 7r + 7r~, and
B° — > 7t 7t°. This asymmetry is often written as

Acp(t, /) = AfCos(Amt) - Sjsm(Amt) (42)

where Aj characterizes the direct CP violation due to interference of different diagrams
contributing to the decay, and Sj measures the indirect CP violation which originates from
mixing between the B° and B° initial states. Measuring the time dependence of the CP
asymmetry allows one to separate the contributions of these two mechanisms.

Similar to the case of the time independent CP asymmetry above, Aj first occurs at order
a s , and can be calculated with knowledge of only the leading order Wilson coefficients. Note
that Aj is simply the time dependent asymmetry evaluated at t — 0, and is given by (|4L)|).
Sj on the other hand is non-zero in naive factorization, and its determination requires
knowledge of the NNLO Wilson coefficients. As such, we will not consider Sj any further
in this paper.

If we write the Feynman amplitudes (|38| ) -(|39| ) in the form A = X^u + X^'c and decompose
the two terms into perturbative contributions

2

U = U + -^-Ui + -—^(3 U2

An (47r y

c = co + g Cl + ^/5 c 2 (43)

then we have

A^ = -2Im[AM(AM)*1 g^j (44)

|Al j | 2 |w| 2 + |A^| 2 |c| 2 + 2Re[Xi'(Xy)*]Re[u*c\

We can expand A^p to order O(a'j.(3o) to obtain

,D ir _ 2Im[A('>(A(')

(A^'^o + A^co)((AS ; )*«o + (A^)*co)
± (uolmfci] - CoImK]) + ^

x \ — (itoIm[ci] - c Im[wi]) + jj^Pq (u Im[c 2 ] - c Im[w 2 ]) \ (45)

16

where we have used the fact that Uq and Co are real. Note that to this order the direct CP
asymmetry is not sensitive to the real part of the perturbative corrections. Note also that, as
anticipated, we require only the next-to-leading order behaviour of the Wilson coefficients.

B. Comparison to Previous Work

As mentioned earlier, the calculations presented in this paper are similar to calculations
presented earlier by Neubert and Pecjak in j^J. Performing a renormalon analysis, they
estimated both perturbative and power corrections in the context of the large-/3o limit. This
limit is a way of organizing the perturbative expansion that differs from what is typically
done in renormalization group improved (RG-improved) perturbation theory. The most
important difference is the power counting. Rather than expanding in the strong coupling,
j3o is taken to be large and one expands in powers of 1//3q. a s is still considered to be a
small parameter in this limit and scales like a s ~ 0(1/ (3q). In practice the large-/?o scaling
is implemented by switching to a rescaled coupling b(fi) related to the leading order running
of a s :

a,M - b(n) = = log J /A 2_y (46)

where b(fi) ~ 0(1)- In contrast to RG-improved perturbation theory, \og(M/fi) is of 0(1)
in the large-/5o limit. Furthermore, before one expands in 1//3q, all occurrences of n/ are
replaced by rif — > — 3/3 /2.

This unusual counting scheme forces one to sum certain classes of diagrams to all orders
in perturbation theory. The fermion bubbles of Fig. Q are a special they scale as

dsfio ~ 0(1) after the replacement rif — > — 3/3q/2. Thus one must sum an infinite number of
fermion bubbles into gluon propagators. The use of such summations is a common technique
in renormalon analyses 0|.

In j^J, in order to calculate the a\ to subleading order, both the hard scattering kernels
and the Wilson coefficients had to be calculated to NLO in the large-/3o limit. As the
hard scattering kernels are 0(1), the Wilson coefficients had to be calculated to O(l/j3 ).
Because they are determined by matching at the weak scale (fi = mw) and running down
to the scale of the decay (/i ~ rrib), it was necessary to have the 0(1/ /3q) matching as well.
However, it was argued by the authors that the difference between the NLO matching and
0(1/ Po) matching was negligible so that the currently known one-loop (NLO) matching was
sufficient.

To run the Wilson coefficients correctly, the elements of the anomalous dimension matrix
had to be determined to the appropriate order. Because the current-current operators enter
at 0(1) in the matching, pieces of the anomalous dimension matrix which depend on these
operators were determined to O(l//3o). Current-current operators affect the running of both
the penguin operators and the current-current operators themselves. For the penguin oper-
ators, their effect can be determined from the LO anomalous dimension matrix in j^]. For

17

the current-current operators this has been calculated in |53[. Penguin operators enter at
order G(1/(3 Q ) at the matching scale. As such, to calculate the elements of the anomalous
dimension matrix which depend on the penguin operators, it was only necessary to calcu-
late the diagrams to 0(1). Unlike the current-current operators, penguin diagrams in the
effective theory can be of 0(1) because of factors of rif which occur in fermion loops.

In contrast, the calculations we perform in this paper are in the context of the usual
RG-improved perturbation theory, where log(M//x) ~ 0(l/a s ), and one calculates order by
order in a s . We have calculated the hard scattering kernels to 0{ol 2 s (3q). This corresponds
to inserting a single fermion bubble into the gluon propagators of the 0(a s ) diagrams and
replacing rif — >• — 3/?o/2. Both the previous authors and ourselves had to decide what to
do about the factors of rif which appear in the penguin diagrams. The factors of rif which
enter from the fermion bubble have corresponding diagrams with gluon and ghost loops
which justify their replacement, but these diagrams are not present for other factors of rif
which emerge from penguin diagrams. The previous authors took two different approaches
to this problem and considered the cases where they either replaced rif — > — 3/3q/2 or they
left these factors of rif alone. They achieved better results from the second approach, which
is physically better justified. We choose to use only this latter approach in this paper.

To further aid in our calculations, we choose to calculate quantities that first occur at
0(a s ) in perturbation theory. It is easily understood that such quantities require only the
NLO Wilson coefficients from j4^|. The leading order terms in the Wilson coefficients sum
logs of the form a™log n (M//^) and scale as 0(1). The NLO terms sum logs of the form
a n+i \ Q g n (M/ fi) and scale as 0(a s ). Because a two loop calculation is necessary to calculate
these NLO terms they may contain factors of (3q in the form

< +2 /3 log n+1 (M//i)~« s /3 . (47)

If we were only after a 0(a s ) result, we would need the 0(1) (LO) Wilson coefficient
and the 0(a s ) hard scattering kernel. At 0(a^/3 o ) however, we need not only the 0(1)
Wilson coefficient and the 0{ol 2 s [3q) hard scattering kernel, but also the 0(a s ) (NLO) Wilson
coefficient and the 0(a s ) hard scattering kernel. Since we only need the piece of the NLO
Wilson coefficient proportional to (3q we are effectively keeping some unnecessary higher
order pieces. Either way, we require only the NLO Wilson coefficients.

An important effect of the differences between the two approaches is reflected in the
different contributions to the hard scattering kernels which must be calculated. In order
to calculate the coefficients (|15jl . we need many hard scattering and vertex contributions.
Neubert and Pecjak needed fewer of these contributions, but those they did need were
needed to all orders in . This difference has important phenomenological effects.

The contributions we include are sensitive to unknown non-perturbative parameters. As we
will see, these parameters can introduce a large uncertainty in various observables.

18

TABLE I: Numerical values of the Wilson coefficients Cj in the NDR scheme at NLO, in units of
10~ 3 . We have used the input parameters Aqcd = 223 MeV, m t = 174 GeV, m b = 4.2 GeV, and
mw = 80.4 GeV. The soft scales are defined using A^ = 500 MeV.

A h m b j2 ^A h m b ^A h 2m b

m b /2

m b

2 m b

Ci

1258.0 1195.2

1150.4

1147.7

1087.8

1048.6

c 2

-474.8 -378.7

-305.3

-300.7

-193.3

-114.4

c 3

35.7 27.4

21.6

21.2

13.8

9.0

c 4

-77.7 -62.8

-52.0

-51.3

-36.0

-25.0

c 5

11.9 12.4

11.9

11.9

9.9

7.7

c 6

-118.6 -88.2

-68.5

-67.4

-43.3

-28.3

°8 9

-169.0

-151.0

-136.0

C. Input Data

The numerical analysis in section II V Dl requires various parameters as theoretical inputs.
In this section we collect these input parameters together.

1. Model Independent Parameters

For the running coupling a s (p) we use

4:71

a s (p)

1

A)ln0 2 /A^ CD

where, in terms of the number of colors N c and flavours rif
llN c -2n f n 34V 2 10iV>

A ln(ln(/x 2 /A^ CD ))
HI ln^/A 2 ^)

0i

- 2C F n f ,

2N r

(48)

(49)

We take Aqcd = 223 MeV, which is equivalent to running with rif — 5 flavours from
a s (M z ) = 0.1185.

The Wilson coefficients calculated to NLO in QCD are shown in Table |U
We choose to work in the Wolfenstein parametrization

/

CKM

1-^

2

A A\ 3 (p - it]) \

A 2

f A\ 2

o(\ 4

(50)

-A 1
\AA 3 (1- p-irj) -AX 2 1 J

In the analysis below we will sometimes plot observables as a function of the unitarity angle
7 = Arg[V^ 6 ]. In that case we write (p — irj) = \J p 2 + rj 2 e~ 11 . We take the numerical values

19

of the CKM parameters from a recent global fit Q: A = 0.83 ± 0.04, A = 0.2224 ± 0.0020,
y/p 2 + r]' 2 = 0.398 ± 0.040, and 7 = (64 ± 11)°.

For the B meson lifetimes we use the PDG values j^: r(B°) = 1.536 ± 0.014ps and
t(B ± ) = 1.671 ± 0.018ps. Our quark pole masses are m^, = 4.2 GeV, m c = 1.3 GeV, and
we set m S7 u,d = 0. Finally, for the 'chiral enhancement' factor defined in ((7|), which is a
renormalization scale dependent quantity, we use

/xp(m 6 /2) = 0.85^, fx P (m b ) = 1.14^, fi P (2m b ) = 1.42^ (51)

for both P = Ti, K.

2. Model Dependent Parameters

The matrix element for a B transition to a pseudoscalar state M is given by

(M(q)\q 7 »b\B(h))

rpB—^M
^0

(P 2

(P 2

m

B

m

M ft

p M (52)

where the momentum transfer is p = h — q. In practice this matrix element is always
contracted with one of the meson momenta, and using the Dirac equation it is always possible

j?B^M
^0

{p 2 ){m

B

m

M)i

so that

to write these contractions in terms of (M(q)\q^>b\B(h))
dependence on F + drops out. Since we are studying mesons with mass small compared to
the B mass, we need consider only the point F^~*^(0). Estimates of this quantity have
been made from QCD light-cone sum rules relativistic quark models |5J|, and

lattice calculations [soj l . with good agreement between the various methods. Numerically
we take

F r K (o)

Jit

F<f ^(0) = 0.258 ± 0.031 GeV.

The decay constants fu are defined by

{M(p)\qY l5 q\0) = -if MP ».

(53)

(54)

In our data analysis we will take

U = 0.1307 ± 0.0004 GeV|55|, f K = 0.1598 ± 0.0016 GeV|55(, f B = 0.180 ±0.040 GeV|6j.

(55)

The general decomposition of the LCDAs $m has been given earlier (fTUj) in terms of
the parameters af 1 , and throughout section IIHI we stated our results in terms of these

parameters. A variety of phenomenological and sum rule estimates have been made for

these parameters |45j, |49j, Ml |£2j , and we adopt the values

a

A

0.

0.10 ±0.12,

(.1;

a

A

0.1 ±0.3,
0.1 ±0.3.

(56)

20

Owing to their nonperturbative origin all of these parameters are rather poorly known, and
this is indicated by the conservative error estimates. The exception to this rule is a\ which
deviates from zero only by SU(2) breaking effects.

Little is known about the LCDA for the light quark in the B meson. Accordingly, in
section llH Bl we. following Ref. |24j] . parametrized the integrals over that we encountered:

^^^-^ r'^^l^^Jjj) (57)

£ A B A h Jo £ \ B A h \m B J

where we take the soft scale to be = 500 MeV. In our numerical analysis we assign a
100% uncertainty to these integrals, varying < ttib/Xb < 2(ms/Aft) and < —uib/Xb <
2(mB/A/, ln(mfi/Ah)) with a uniform probability distribution.

Also in section IIIIBI we saw that integrals over the twist-3 LCDAs required the intro-
duction of two other parameters

dy-2-^-=Xg\ / dy » I »' =Xp. (58)

o y Jo y

The approximate magnitude of these parameters can be estimated by power counting, but
in general they can be complex. Therefore, following Ref.

24 we write them as

= (1 + PH e^) In (^) , X? = (1 + p H <**) Li 2 {-^) . (59)

In the numerical analysis below we vary < pn, Ph < 2 and allow the phases 4>h, 4>h to
take arbitrary values.

In the numerical analysis which follows, our central values are obtained by setting all of
the input parameters at the center of their ranges. For the arbitrary phases in Xh and Xh
we must choose a particular value for our 'central value'. We choose (frHi &H — 7r/2, which
makes the real and imaginary parts of the central values of Xh and Xh of equal magnitude.

D. Results

The main numerical results of this paper are presented in Table HT1 which shows the results
for the direct CP asymmetries A^p. After a general discussion of the results, we include
a more detailed discussion of the particular asymmetry A^(ti + 7t~). For each quantity we
first state the prediction at 0(a s ), then the O(a^/3 ) correction and, finally, the sum. We
also state the results at three different renormalization scales.

In addition, we estimate the uncertainties arising from the model dependent parameters
discussed in section IIV C 21 We divide these parameters into three groups. The Gegenbauer
moments which parametrize the shape of the LCDAs, are varied with the 1 a error bars
given in (J56j) . The B decay formfactor ^^^(O) and B meson decay constant are varied
with the 1 a error bars given in (J53|) and (J55j) . respectively. Finally, the parameters arising

21

from the spectator scattering graph (Xb, Xb, Xh, Xh) are varied with equal probability over
the ranges given in section Q V C 21 above. These three groups are labelled LCDA (c^f), FF
(f B , Fq"* 7 "^)), and SPEC (A#, Xb, X H , X H ) in our Table. The sets of parameters are varied
independently, and in all cases the uncertainties we give are 1 a standard deviations. Note
that there are additional sources of uncertainty we do not consider, such as dependence on
the CKM matrix elements and quark masses. Our analysis does, however, give insight into
the relative size of perturbative corrections and nonperturbative uncertainties.

1. CP asymmetries

Table |H] contains our results for the direct CP asymmetries A^p. It should be noted
that the values in Table ITT1 do not take into account the contribution from weak annihilation
diagrams and as such should not be taken as rigorous predictions of the BBNS method.
They are however valid for their purpose of studying the perturbative behaviour of the
formalism. Notice that the asymmetry Acp{^~^°) is not shown in the table; as is clear
from the definitions (}39|) . the amplitude for B~ — > 7t~tt° has only one weak phase and
therefore the asymmetry for this mode is zero (up to small electroweak corrections which
we have neglected). The final column of Table HP gives the current experimental values for
the CP asymmetries. They are from HFAG, Summer 2005 compilation jf^j], apart from the
observables where different experiments do not agree, in which case the errors are inflated
according to the PDG prescription |55|.

Though the relative sizes of the perturbative contributions at 0(a s ) and O(a1/3o) in Table
Clare quite sensitive to the renormalization scale, it is generally true that the two contribu-
tions are of roughly the same size. This may be understood as follows: the asymmetries are
dominated by the contributions from the penguin diagrams, and a s (3o\P / P\/ Ait ~ 1, where
P and P refer to the penguin functions defined in section IIII CI

There is no reduction in the renormalization scale dependence of the asymmetries after
adding the O(al/3o) terms. This behaviour follows from our previous remarks: at a given
scale, the O(a^j3 ) contributions are numerically similar to the 0(a s ) contributions. The
sum, therefore, follows the pattern established at 0(a s ).

The next to last columns of Table |H] show the sensitivity of the asymmetries to the three
classes of parameters defined above. The dominant uncertainty for most of the modes is due
to the light cone distribution amplitudes (LCDA). These parameters are similar in size for
each decay mode (see (jSHJ)), and the asymmetry is proportional to them. Consequently the
size of the uncertainty in this column scales roughly with the size of the asymmetry itself.

Note also that three of the modes - 7i°K~, n°Ks, and 7r°7r° - are particularly sensitive
to the form factor (FF) and spectator scattering (SPEC) parameters. Unlike other decay
modes, the FF and SPEC parameters in these modes are proportional to the Wilson coeffi-
cient C\. The strong sensitivity in these modes is simply a reflection of the large size of C\

22

TABLE II: Numerical results for direct CP asymmetries A^p, expressed in percent. We present the
0{a s ) and subleading O{a 2 s (3o) perturbative corrections. Partial error estimates whose meaning is
explained in the text and experimental values are also presented.

Decay Mode

H 0(a s ) O(a 2 s p )

Total

Error Estimates
LCDA FF SPEC

Experiment

Acp(tt-K°)

m b /2 0.9 0.1
m b 1.0 0.3
2m b 1.2 0.4

0.9
1.2
1.6

±0.1 ±0.0 ±0.0
±0.1 ±0.0 ±0.0
±0.1 ±0.0 ±0.0

-2. ±5.

Acp(tt°K~)

m b /2 9.1 -1.4
m b 15.9 8.3
2m b 29.4 28.9

7.7
24.2
58.2

±2.5 ±0.5 ±1.4
±3.0 t° 6 7 ±1.8
±4.1 ±oi ±2.5

4. ±4.

A C p(tt + K-)

m b /2 3.9 -3.1
m b 9.3 4.8
2m b 19.2 20.1

0.7
14.0
39.3

±2.2 ±0.0 ±0.1
±2.3 ±0.0 ±0.0
±3.1 ±0.0 ±0.1

-11.5 ± 1.8

Acp(n K s )

m b /2 -3.0 -1.2
m b -3.2 -2.0
2m b -4.5 -4.3

-4.2
-5.2
-8.8

±0.9 ±0.4 ±1.1
±1.0 ±0.4 ±1.2
±1.1 t°j ±1.4

2. ± 13.

A C p(TT + ir~)

m b /2 -2.6 1.6
m b —3.6 —1.9
2m b -4.2 -4.3

-1.0
-5.6
-8.5

±1.1 ±0.0 ±0.1
±0.7 ±0.0 ±0.0
±0.5 ±0.0 ±0.0

37. ± 23.

Acp(k°k )

m b /2 68.6 28.8
m b 40.7 26.5
2m b 25.3 25.4

97.4
67.2
50.7

±16.0 ±\%l ±33.5
±10.4 ±19.2
±5.7 ±Ia ±9.5

28. ± 40.

in comparison to the other Wilson coefficients.

On the other hand, the asymmetry .A^p (7r~.fr ) has no dependence on the form factors or
spectator scattering parameters. This may be understood by examining our master formula
for the CP asymmetry, Eq. (j45j) .

^cp K Im K °\ = IT ( M oIm[ci] - colmfui]) + tj-t^A) (u lm[c 2 } - c Im[u 2 ]) . (60)

47T [4:71 ) z

Because of the particular form of the amplitude for B~ — > n~K° shown in (jSHJ), = Co
and Im[wj] differs from Im[cj] only by QCD penguin contributions. The result is that in ()60|)
only the QCD penguins contribute to the asymmetry. Consequently this mode is insensitive
to most of the model dependence in the BBNS framework.

It is important to note that our values of the CP asymmetries at 0(a s ) for B~ — > ir°K~
and B° — > ir + K~ exhibit a much greater scale dependence than those of Beneke and Neubert

23

|64| • in our calculation of these asymmetries we keep the real parts of our amplitudes to only
LO in a s and it is the large scale dependence of these real parts that leads to the large scale
dependence of our asymmetries. Beneke and Neubert on the other hand keep the real parts
of their amplitudes to NLO and it is these higher order terms that result in their reduced
scale dependence.

Because of their different focus, it is difficult to compare our results to those of Neu-
bert and Pecjak j^J. These authors were primarily interested in estimating the size of
non-perturbative corrections. Although they did calculate some leading perturbative cor-
rections, they sought only to compare the size of these corrections to their estimate of the
power corrections. As such, almost all of the parameters we chose to vary, including the
renormalization scale, the LCDAs, as well as the form factors and decay constants, they
simply held fixed, and no estimate of their induced uncertainties was made. They did calcu-
late the CP asymmetries for the B° — > tt + K~ and B~ — > ir°K~ decay modes. Their results
are consistent with our own, to within our large uncertainties. Perhaps the most significant
comparison concerns the size of the subleading corrections. They found, as we did, that the
subleading corrections are substantial and can be almost as large as the leading order result.

Recently new measurements of the CP asymmetry in B° — > 7t + tt~ were released by the
BABAR and BELLE Q collaborations. In Figure El we show our results for the direct

CP violation parameter A nn defined in ()42)1 as a function of the unitarity angle 7. The
current experimental data for this quantity is

BABAR l3 : A n7T = 0.09 ± 0.15(stat) ± 0.04(syst)
BELLE0 : A n7T = 0.56 ± 0.12(stat) ± 0.06(syst), (61)

The la ranges for these measurements are superimposed in Figure El

Figure El shows that the calculated CP asymmetry has a large renormalization scale
dependence at O(a^(3o), which dominates the uncertainty in the prediction. Within the large
error bars, the experimental results of BABAR are in fair agreement with these calculations,
while the results of BELLE show a several a deviation. Clearly, more work is required on
both the experimental and theoretical sides before any definitive statement can be made
about the success of the BBNS framework in this context. For instance, the contributions
from power-suppressed annihilation diagrams should be included, as they are known to have
a large effect on strong phases 0, .

V. CONCLUSIONS

In this paper we have calculated perturbative corrections to B — > irn, irK decays up to
O(a 2 s (3o) in the QCD factorization formalism, including 'chirally enhanced' power correc-
tions but neglecting other corrections entering formally at (^(AqcdA 71 !))- We have included
contributions from non-factorizable vertex diagrams, QCD penguin diagrams, and spectator

21

0.6-

0.4

25 50 75 100 125 150 175
y (degrees)

FIG. 5: The CP violating quantity A nn as a function of the unitarity angle 7. The three short-
dashed curves are the prediction at order 0(a s ), while the three solid curves include the perturba-
tive corrections up to O(a%/3o). The lines in each set correspond to the three different renormal-
ization scales fi = m&/2, fi = nib and /i = 2rrib- The heavy dashed and dot-dashed horizontal lines
are the la experimental uncertainties for BABAR and BELLE, respectively.

scattering diagrams. In all cases we have derived analytic expressions for the hard scattering
kernels for general light-cone quark momentum distribution functions.

We have used these analytic results to study the direct CP asymmetries for a set of
phenomenologically interesting decay modes. We focused on the behaviour of perturbation
theory for this observable, and we estimated the uncertainties due to model dependent
parameters.

For the direct CP asymmetries A^p, we found that contributions at 0(ol 2 s (3q) are compa-
rable to those at 0(a s ) for all the modes. This conclusion is in agreement with the results
of jl^l , which indicated a large perturbative correction between one- loop and two-loop order
in the large- (3q limit. As well, we found a very strong dependence on the renormalization
scale; in some cases the asymmetry varies over an order of magnitude.

For all modes, with the exception of A^p(ir 7T ), the primary uncertainty at a given scale
arises from uncertainty over the shape of the light cone momentum distribution amplitude.
The uncertainties arising from form factors and spectator scattering model parameters are
mode dependent and relatively small. The asymmetry Aqp(7i~T?°) is particularly clean
in the QCD factorization framework, having no dependence on form factors or spectator
scattering parameters. Finally, we have examined the direct CP violation parameter A n7T
in the B° — > ti + tx~ decay mode and have found a large perturbative correction at 0{ol 2 s (3q).
The result agrees with the current experimental measurements, though the errors for both
theory and experiment are large.

25

VI. ACKNOWLEDGMENTS

We would like to thank Michael Luke for discussions related to this project. This work
is supported by the Natural Sciences and Engineering Research Council of Canada and by
the Walter B. Sumner Foundation.

D. Cronin-Hennessy et al. [CLEO Collaboration], Phys. Rev. Lett. 85, 515 (2000).

D. M. Asner et al. [CLEO Collaboration], Phys. Rev. D 65, 031103 (2002)

|arXiv:hep-ex/0103040| .

S. Chen et al. [CLEO Collaboration], Phys. Rev. Lett. 85, 525 (2000) arXiv:hep-cx/0001009 .
B. Aubert et al. [BABAR Collaboration], Phys. Rev. Lett. 87, 151802 (2001)
|arXiv:hep-ex/0105061| .

B. Aubert et al. [BABAR Collaboration], Phys. Rev. D 65, 051502 (2002)
|arXiv:hep-ex/0110062| .

B. Aubert et al. [BABAR Collaboration] , |arXiv:hep-ex/0109007l
B. Aubert et al. [BABAR Collaboration], |arXiv:hep-ex/0 206053 .

B. Aubert et al. [BABAR Collaboration], Phys. Rev. Lett. 89, 281802 (2002)
|arXiv:hep-ex/0207055| .

B. Aubert et al. [BABAR Collaboration] , |arXiv:hep-ex/0207063|
B. Aubert et al. [BABAR Collaboration] , |arXiv:hep-ex/0207065|

B. Aubert et al. [BABAR Collaboration], Phys. Rev. Lett. 93, 231804 (2004)
|arXiv:hep-ex/0408017| .

B. Aubert et al. [BaBar Collaboration], |arXiv:hep-ex/ 0501071

K. Abe et al. [BELLE Collaboration], Phys. Rev. Lett. 87, 101801 (2001)
|arXiv:hep-ex/0104030| .

K. Abe et al. [BELLE Collaboration], Phys. Rev. D 64, 071101 (2001) |arXiv:hep-ex/0106095| .
K. Abe et al. [Belle Collaboration], Phys. Rev. Lett. 93, 021601 (2004) |arXiv:hep-ex/0401029| .
B. C. K. Casey et al. [Belle Collaboration], Phys. Rev. D 66, 092002 (2002)
|arXiv:hep-ex/0207090| .

K. Abe et al. [Belle Collaboration], |arXiv:hep-ex/ 0502035

M. Bauer, B. Stech and M. Wirbel, Z. Phys. C 34, 103 (1987).

A. Deandrea, N. Di Bartolomeo, R. Gatto and G. Nardulli, Phys. Lett. B 318, 549 (1993)
|arXiv:hep-ph/9308210| .

M. Neubert, V. Rieckert, B. Stech and Q.P. Xu, in: Heavy Flavours, ed. A.J. Buras and M.
Lindner (World Scientific, Singapore, 1992), pp. 286.

M. Neubert and B. Stech, Adv. Ser. Direct. High Energy Phys. 15, 294 (1998)

|arXiv:hep-ph/9705292 .

26

M. Beneke, G. Buchalla, M. Neubert and C. T. Sachrajda, Phys. Rev. Lett. 83, 1914 (1999)
|arXiv:hep-ph/9905312| .

M. Beneke, G. Buchalla, M. Neubert and C. T. Sachrajda, Nucl. Phys. B 591, 313 (2000)
|arXiv:hep-ph/0006124| .

M. Beneke, G. Buchalla, M. Neubert and C. T. Sachrajda, Nucl. Phys. B 606, 245 (2001)
|arXiv:hep-ph/0104110| .

C. W. Bauer, D. Pirjol and I. W. Stewart, Phys. Rev. D 65, 054022 (2002)
|arXiv:hep-ph/0109045| .

C. W. Bauer, D. Pirjol and I. W. Stewart, Phys. Rev. Lett. 87, 201806 (2001)

arXiv:hep-ph/0107002|.

C. W. Bauer, D. Pirjol and I. W. Stewart, Phys. Rev. D 67, 071502 (2003)
|arXiv:hep-ph/0211069| .

J. Chay and C. Kim, Phys. Rev. D 65, 114016 (2002) |arXiv:hep-ph/0201197| .
J. g. Chay and C. Kim, |arXiv:hep-ph/0205117l

M. Beneke, A. P. Chapovsky, M. Diehl and T. Feldmann, Nucl. Phys. B 643, 431 (2002)
|arXiv:hep-ph/0206152l .

I. Z. Rothstein, Phys. Rev. D 70 (2004) 054024 |arXiv:hep-ph/0301240| .

J. Chay and C. Kim, Nucl. Phys. B 680, 302 (2004) |arXiv:hep-ph/0301262| .

C. W. Bauer, D. Pirjol, I. Z. Rothstein and I. W. Stewart, Phys. Rev. D 70, 054015 (2004)
|arXiv:hep-ph/0401188| .

T. Muta, A. Sugamoto, M. Z. Yang and Y. D. Yang, Phys. Rev. D 62, 094020 (2000)
|arXiv:hep-ph/0006022| .

D. s. Du, D. s. Yang and G. h. Zhu, Phys. Lett. B 488, 46 (2000) |arXiv:hep-ph/0005006| .

D. s. Du, H. u. Gong, J. f. Sun, D. s. Yang and G. h. Zhu, Phys. Rev. D 65, 074001 (2002)
|arXiv:hep-ph/0108141| .

K. G. Chetyrkin, A. L. Kataev and F. V. Tkachov, Phys. Lett. B 85, 277 (1979);

M. Dine and J. R. Sapirstein, Phys. Rev. Lett. 43, 668 (1979);

W. Celmaster and R. J. Gonsalves, Phys. Rev. Lett. 44, 560 (1980).

S. Narison and A. Pich, Phys. Lett. B 211, 183 (1988);

E. Braaten, Phys. Rev. D 39, 1458 (1989).

M. E. Luke, M. J. Savage and M. B. Wise, Phys. Lett. B 343, 329 (1995)
|arXiv:hep-ph/9409287| ;

T. van Ritbergen, Phys. Lett. B 454, 353 (1999) |arXiv:hep-p"h"79 903226 ;

M. Steinhauser and T. Seidensticker, Phys. Lett. B 467, 271 (1999) |arXiv:hep-p h /9909436 .

M. Neubert and B. D. Pecjak, JHEP 0202, 028 (2002) |arXiv:hep-ph/0202128| .

C. N. Burrell and A. R. Williamson, Phys. Rev. D 64, 034009 (2001) |arXiv:hep-ph/0101190| .

T. Becher, M. Neubert and B. D. Pecjak, Nucl. Phys. B 619, 538 (2001)

|arXiv:hep-ph/0102219| .

27

[43] G. Buchalla, A. J. Buras and M. E. Lautenbacher, Rev. Mod. Phys. 68, 1125 (1996)
arXiv:hep-ph/9512380] .

A. V. Manohar and M. B. Wise, Phys. Lett. B 344, 407 (1995) |arXiv:hep-ph/ 9406392 .

V. M. Braun and I. E. Filyanov, Z. Phys. C 44, 157 (1989) [Sov. J. Nucl. Phys. 50, 511.1989

YAFIA,50,818 (1989)].

V. M. Braun and I. E. Filyanov, Z. Phys. C 48, 239 (1990) [Sov. J. Nucl. Phys. 52, 126
(1990)].

M. Beneke, Nucl. Phys. Proc. Suppl. Ill, 62 (2002) |arXiv:hep-ph/0202056| .
G. P. Lepage and S. J. Brodsky, Phys. Rev. D 22, 2157 (1980).
V. L. Chernyak and A. R. Zhitnitsky, Phys. Rept. 112, 173 (1984).

S.V. Mikhailov and A.V. Radyushkin, JETP Lett. 43 712 (1986); Sov. J. Nucl. Phys. 49 494
(1989); Phys. Rev. D 45 (1992) 1754.

A. J. Buras and P. H. Weisz, Nucl. Phys. B 333, 66 (1990).
M. Beneke, Phys. Rept. 317, 1 (1999) |arXiv:hep-ph/9807443| .
N. Pott, | arXiv:hep-ph/9710503|
A. Ali, |arXiv:hep-ph/0312303|

S. Eidelman et al. [Particle Data Group], Phys. Lett. B 592, 1 (2004).

A. Khodjamirian, R. Ruckl, S. Weinzierl, C. W. Winhart and O. I. Yakovlev, Phys. Rev. D
62, 114002 (2000) |arXiv:hep-ph/0001297| .

P. Ball and R. Zwicky, JHEP 0110, 019 (2001) |arXiv:hep-ph/0110115| .

P. Ball and R. Zwicky, Phys. Rev. D 71, 014015 (2005) |arXiv:hep-ph/0406232| .

D. Melikhov and B. Stech, Phys. Rev. D 62, 014006 (2000) |arXiv:hep-ph/0001113| .

A. Abada, D. Becirevic, P. Boucaud, J. P. Leroy, V. Lubicz and F. Mescia, Nucl. Phys. B

619, 565 (2001) |arXiv:hep-lat/0011065| .

A. Abada, D. Becirevic, P. Boucaud, J. P. Leroy, V. Lubicz, G. Martinelli and F. Mescia,

Nucl. Phys. Proc. Suppl. 83, 268 (2000) |arXiv:hep-lat/9910021| .

V. M. Braun and A. Lenz, Phys. Rev. D 70, 074020 (2004) |arXiv:hep-ph/0407282| .

Heavy Flavor Averaging Group (HFAG) data averages can be found at

www . slac . st anfor d . edu /xor g/hf ag /

M. Beneke and M. Neubert, Nucl. Phys. B 675, 333 (2003) |arXiv:hep-ph/0308039| .

Y. Y. Keum, H. n. Li and A. I. Sanda, Phys. Lett. B 504, 6 (2001) |arXiv:hep-ph/0004004| .